Information on Neca's Rich Representation Language

This directory holds deliverables, technical documents and demos related to the ongoing design of NECA's internal representation language.

----------------------- RRL related deliverables ------------------------
D3a: Specification of Scene Descriptions for the Neca Domain
Deliverable by Paul Piwek, ITRI. April 2002.
Full paper (PDF)

D9c: Report on the assessment of existing markup languages for avatars, multimedia and multimodal systems on the WWW
Deliverable by Hannes Pirker and Brigitte Krenn, OFAI. May 2002.
Full paper (270k - PDF)

----------------------- XML Schemas ------------------------
rrl main schema
Version 0.4, Feb 2003.

domain dependent schema: Socialite
Version 0.4, Feb 2003.

domain dependent schema: eShowRoom
Version 0.4, Feb 2003.

Auxiliary scheme for "mimicing" SMIL
Version 0.1, December 2002.

----------------------- RRL Examples ------------------------
The RRL of a Socialite Dialogue throughout differen processing steps
necaRRL 0.4, Feb 2003
Step 1: The RRL which is passed to the speech synthesis module

Step 2: The RRL output of the speech synthesis: speech timing information is added

Step 3: The RRL output of the gesture assignment module: Viseme information and temporal alignment is added. This RRL subsequently will be transformed to a player-specific format (i.e. flash, Microsoft Agent)

----------------------- Gesticon ------------------------
The Gesticon i.e. the lexicon of gesture is the central repository for all information needed in the context of assigning and selecting gestures throughout the processing in NECA, and is tightly coupled to the RRL.

Gesticon used by the eShowRoom

XML schema for the gesticon

----------------------- Discussion Notes ------------------------
Version 0.4, March 2003.
Originating from an internal discussion text, this document comes closest to a "tutorial" on RRL. Includes explanation on the elements and attibutes used and contains examples.
Technical Note (83k - PDF)

----------------------- RRL related Papers and Presentations ----------
RRL: A Rich Representation Language for the Description of Agent Behaviour in NECA
Piwek P., Krenn B., Schröder M., Grice M., Baumann S., Pirker H., In Proceedings of the Workshop ``Embodied conversational agents - let's specify and evaluate them!'', held in conjunction with AAMAS-02, July 16 2002, Bologna, Italy
Full paper (235k - PDF)
Slide show (1.1 MB - Powerpoint)

The Rich Representation Language (RRL) is an XML compliant language for the representation of agent behaviour at various levels of abstraction. It focuses on representations of agent behaviour in dialogue and supports the integrated representation of various types of information (pragmatic, semantic, syntactic, prosodic, etc). The language builds on various existing proposals and standards (SMIL, MURML, etc.). It is being deployed as an interface language in which the modules of NECA system exchange information.